what is the strange stuff at top left of plot_convergence

I make a gp_minimize with 20 calls : the correct values of the functions are the blue bottom markers.

What is the strange markers distributions at the top left ?

optimize_result=gp_minimize(func_objective,space_xgboost,n_calls=n_calls,n_initial_points=n_initial_points,x0=x0_init,random_state=1)

ax=plot_convergence(optimize_result,true_minimum=optimize_result.fun)
ax.figure.tight_layout()
ax.figure.savefig(fresults/convergence_hpo_skopt.png)